One of the most common malignant tumors in humans , colorectal cancer has been extensively studied during the past few decades. Staging colorectal cancer allows clinicians to obtain precise prognostic information and apply specific treatment regimens. The American Joint Committee on Cancer (AJCC)/International Union Against Cancer (UICC) tumor-node-metastasis (TNM) staging system serves as a useful tool for prediction of survival in colorectal cancer patients. Metastases, depth of tumor infiltration and lymph node involvement have traditionally been recognized as the most important factors predicting outcome [2,3]. Nevertheless, TNM staging occasionally fails in everyday clinical practice, as some patients in the same clinicopathological stage may exhibit noteworthy variation in outcomes , having greater recurrence and mortality rates than those justified by their TNM stage [4,5]. The fact that approximately 25% of early-stage colorectal cancer patients end up with distant metastases [6,7] has led to the hypothesis that cancer development and progression might depend partly on changes in several histological features with which we were not previously familiar. These previously unrecognized features are closely related to the way cancerous cells interact with the surrounding stroma and obtain their potential for invasiveness . They include tumor budding (TB), poorly differentiated clusters (PDCs), extramural vascular (vein) invasion (EMVI), perineural invasion (PNI), tumor deposits (TDs), and mucin pools (MPs) (Tables 1, 2).
Table 1 Frequency of appearance of extratumoral features in patients with colorectal cancer
Table 2 The impact of extratumoral features on the frequency of nodule invasion, local recurrence and survival rate
Originally described by Imai in the Japanese literature  as a “sprouting” at the invasive front of several carcinomas and subsequently defined by Jass in 1987 , TB represents both a morphological and a histological feature, which generally reflects the detachment of malignant tumor cells from the main tumor mass, resulting in the presence of individual or small clusters of up to five dedifferentiated cells within the peritumoral stroma [11,12]. This phenomenon, seen in approximately 40% of colorectal cancers [13,14] and increasingly detected in advanced TNM stages  and poorly differentiated tumors [12,16], is thought to be the consequence of the loss of adhesion of the tumor cells that produces cell migration at the invasive edge of the tumor, after disruption of the epithelial cell capsule [7,16] (Table 1). Budding cells then invade and disrupt basic membrane and muscularis mucosae, thus allowing the development of metastases though lymphovascular invasion . Interestingly, it seems that invasion by TB cells is facilitated by the arrest of the cell cycle, which allows cancer cells to focus on signals triggering local invasion rather than promoting proliferation .
Arguably , TB cells seem to obtain their growth potential through the process of epithelial-mesenchymal transition (EMT), believed to be a key process in the metastatic cascade [16,20,21]. EMT is described as a process during which a normal epithelial cell is transforming into a spindle cell, elongated and developing pseudopodia. This mesenchymal transformation results in a loss of expression of physiological membrane markers and an increase in invasiveness via elimination of membrane E-cadherin-mediated cell adhesion . This allows cancer cells to detach from the main tumor mass and form buds into the peritumoral connective tissue [13,23]. The reduction in E-cadherin activity is closely related to the nuclear accumulation of beta-catenin, which prevents membrane E-cadherin functioning as a promoter of the cell-to-cell attachment process in TD cells [24,25]. Nuclear β-catenin accumulation has been linked with the risk of nodal involvement and poor prognosis in patients with colorectal cancers related to chromosomal (APC gene) mutation and Wnt-signaling pathway dysfunction [26,27]. On the other hand, nuclear β-catenin expression is rarely seen in high microsatellite instability (MSI-high) colorectal cancers . As a result, membrane E-cadherin function and cell-to-cell adhesion are preserved; thus, MSI-high tumors are significantly less common to exhibit TB [29,30].
TB can easily be detected using conventional hematoxylin-eosin stains at high magnification [23,31]. However, because tumor buds can be obscured by inflammation, immunohistochemical methods, such as pan-cytokeratin staining using anti-cytokeratin antibodies, are needed to properly identify budding cells [19,31,32]. Given adequate experience, interobserver agreement in detecting TB and EMT has been reported to range between 83-96% [31,33,34], although diagnostic methods need further standardization.
In clinical terms, the presence of TB has been proved to increase the risk of lymph node metastases, either in patients with malignant polyps or in those with proven stage II colorectal cancer, thus mandating changes in therapeutic practice . In fact, there is a great body of evidence supporting the role of TB in predicting colorectal cancer aggressiveness, as it seems to be related with poorly differentiated tumors that have a roughly invasive front, which in turn gives rise to vascular and lymphatic invasion [35-37]. It is currently known that, among other factors, TB is strongly correlated with a 9.8-27% risk of lymph node metastases, in either stage I (early T1/2 N0 tumors) [14,26,35,38-44] or stage II (T3/T4 N0 tumors) [14,34,45] colorectal cancers. On the other hand, the number of tumor buds is related to the actual risk of development of lymph node metastases, since the presence of 10 or more tumor buds (high-grade budding)  seems to be a powerful predictor of lymph node involvement [44,46]. High-grade budding has been increasingly detected in poorly differentiated, serrated, BRAF-mutated colorectal tumors , which without exception present with poor prognosis.
The depth of infiltration by budding cells is currently shown to influence the frequency of lymph node metastases. Submucosal invasion <1000 μm (1 mm) is related with a low risk for nodal metastases [35,40-42], irrespective of the budding grade. However, the risk of nodal involvement in tumors invading beyond 1000 μm into submucosa (≥sm2), is related to the grade of TB, being exceptionally high in high-grade budding . Following invasion of the submucosa, budding cells invade the muscularis propria either horizontally, spreading between circular and longitudinal muscle layers (expanding type), or “streaming” into the circular muscle layer (infiltrating type) [8,48]. Either invasion process is related to an increase in the risk of vascular and lymphatic infiltration . Moreover, the horizontal spread of budding cells between circular and longitudinal muscle layers may facilitate perineural invasion in the myenteric plexus, thus providing an alternative way for the tumor to metastasize . Similarly, tumor buds diffusely invading into the circular muscle layer could trigger stromal fibrosis and bowel wall shrinking. Indeed, obstructing carcinomas often present with high-grade infiltrating-type TB .
It has been repeatedly shown that the presence of severe budding increases the risk of locoregional recurrence and lowers the rates of both disease-free (DFS) and overall (OS) patient survival [14,34,45,50,51]. In fact, high-grade budding is associated with lower cancer-specific 5-year survival rates (ranging between 39% and 67%) compared with low-grade budding, related with survival rates ranging between 80% and 92% [34,52]. Similarly, cumulative 5-year survival is inversely related to the grade of budding (93.9% in low vs. 73.9% in high-grade TB patients) .
TB-positive early colorectal cancer patients may present with nodal metastases and stage II colorectal cancer patients with high-grade TB experience similar or even worse outcomes and mortality rates compared to those with TB-negative stage III colorectal cancer [15,23,53]. As a consequence, changes in therapeutic strategies must be strongly recommended. Local excision of a seemingly early-stage tumor and omission of adjuvant chemotherapy for an otherwise stage II cancer may be not oncologically sufficient [34,38,40,44-46,50]. On the contrary, detection of TB does not provide any additional information regarding outcome in node-positive stage III colorectal cancer patients, as they are almost always suitable candidates for (neo)-adjuvant (radio)-chemotherapy , although poor response rates should be anticipated in the presence of severe TB [55,56].
Hypoperfusion of the TB area could be partly responsible for the resistance of high-grade budding tumors to chemotherapy [55,56], as peritumoral stromal vessels surrounding tumor buds have a small perimeter and reduced lumen area compared to vessels located within the tumor center  and are thus susceptible to hypoxia .
Host immune defense mechanisms seem to play a significant role in reducing the detrimental effect of TB . Peritumoral stromal infiltration by inflammatory cells has been found to limit the invasion of budding cells via cell-to-cell interaction, resulting in the immune destruction of the budding clusters [13,59]. This pro-/anti-tumor factors model seems to reliably predict outcomes in colorectal cancer patients, since patients with high-grade TB tumors have a considerably better prognosis under the presence of peritumoral inflammation, exhibiting lower rates of nodal involvement and better survival [13,59-61].
Apart from TB found in the periphery of the tumor mass in resection specimens, it is not unusual for tumor buds to be discovered within the tumor mass itself during preoperative colon and non-radiated rectum biopsies . This so-called intratumoral budding is seen in approximately 17-20% of biopsy specimens . The presence of intratumoral buds in colorectal cancer biopsies is actually related to the development of an aggressive invasive front, as indicated by the increase in peritumoral budding counts found in concomitant surgical specimens. That is, the more severe the intratumoral budding, the higher the grade of peritumoral budding. Intratumoral budding directly reflects the tumor’s biological aggressiveness [33,34,38,40,44-46,50], being associated with a significant reduction in DFS (33% vs. 78%), cancer-specific 5-year survival (61% vs. 87%) and chemotherapy response rates  (Tables 1, 2).
PDCs represent, by definition, groups of five or more dedifferentiated tumor cells with minimal or no glandular formation located at the peritumoral matrix close to the invasive edge [63,64]. It has been suggested that the formation of PDCs represents a sequential step in the growth process of the colorectal cancer, as they seem to develop through the formation of TB [64-66]. Indeed, PDC and TB share similar morphology, both appear at the tumor’s invasive front and are linked to the EMT process .
PDCs are easily detected by the use of hematoxylin-eosin staining at the advancing edge of the tumor, with tumor cells being counted in the microscopic field under a ×20 objective lens [64,67,68]. As PDC foci are larger than TB foci, unlike the latter they are easily identified without the application of cytokeratin-based immunohistochemistry . According to the number of PDC foci identified at the part of the tumor’s invasive site with the highest number of PDCs, a grading system has been developed to categorize colorectal cancer into three grades of malignancy. Grade 1 includes tumors with less than 5 PDC foci, grade 2 includes tumors with 5-9 PDC foci, and grade 3 includes tumors with 10 or more PDC foci [64,67,69,70]. It seems that the number of PDC foci provides higher accuracy and better reproducibility in colorectal cancer staging than the conventional TNM system [71,72].
Up to 40% of colorectal cancers may present with PDCs (Table 1). The incidence increases with the depth of tumor infiltration. Tumors invading beyond 1000 μm into submucosa carry the highest risk. Rectal cancers seem to present preferentially with a PDC pattern [63,64].
It has been suggested that, like TB, the development of PDCs is closely related to Wnt-signaling pathway dysfunction as a result of chromosomal mutation and is therefore associated with failure of the cell-to-cell adhesion process due to loss of membrane E-cadherin . Furthermore, K-RAS mutations seem to be involved in the PDC formation process, as they are significantly related to PDC grade .
PDCs and the micropapillary pattern seen in several colorectal cancers are suspected to share the same biological basis  and occasionally display morphologic similarity. Indeed, although uncommonly seen in approximately 4-13% of colorectal cancers, a micropapillary component may be present over between 5-95% of their surface, located mostly at their invasive front [74,75]. The micropapillary component consists of clusters of cancer cells located within intra- or peritumoral empty spaces resembling (but not being) lymphatic vessels, lack a fibrovascular core and present reverse polarity, resulting in secretory activity in the stroma-facing surface . When compared to those without, colorectal cancers with a micropapillary component exhibit aggressive biological behavior, as they have greater lymphatic metastatic potential [74,76], a higher risk of being unresectable  and significantly lower survival rates .
Several studies have documented the impact of PDCs on the risk of lymph node metastases and patient survival. The development of PDCs is undeniably related to lymphatic infiltration and lymph node metastases [64,67,68], as well as occult lymph node micrometastases in otherwise node-negative colorectal cancers [63,68]. The PDC pattern is also associated with perineural invasion [64,67,77] and EMVI . As a result, both, DFS and OS are greatly influenced by the presence of PDCs, being inversely related to the number of PDC foci [64,67,69,70,72,77]. Thus, the presence of PDCs may call for upstaging of stage I/II colorectal cancer patients, as many of them exhibit similar or even worse survival rates compared to stage III patients [68,72]. The grade of the PDC pattern also has a significant impact on the tumor’s response to chemotherapy, as colorectal cancers with more than 10 PDC foci (grade 3 tumors), especially metastatic ones, show no significant post-chemotherapy benefit  (Tables 1, 2).
EMVI invasion, predominately venous, is thought to be a distinct predisposing factor leading to poor outcomes in colorectal cancer patients. It is defined as a rounded mass of tumor cells located in an endothelium-lined space surrounded by a smooth muscle layer and/or contains red blood cells . Gastrointestinal pathologists are nowadays able to indirectly suspect or even diagnose the presence of venous invasion by detecting either a tumor mass, near or adjacent to an artery, without identification of an accompanying vein (“orphan” artery sign), or the presence of smooth tongues of tumor in the surrounding fat, the so-called “protruding tongue” sign [3,79]. However, diagnosing EMVI can occasionally prove very challenging, especially in the hands of non-specialized pathologists; thus, the reported incidence varies widely between 11% and 90%, suggesting that EMVI is underreported worldwide [79-81] (Table 1). The use of elastin-staining techniques that differentiate veins from lymphatic vessels by highlighting the elastic fibers in the wall of the former, have added significant advantage in diagnosing vascular invasion by increasing the sensitivity of detecting EMVI more than twofold [79,82,83], thus improving interobserver agreement significantly. Immunohistochemistry often fails to detect EMVI, as the endothelium of the affected veins is frequently completely destroyed.
Apart from the biopsy specimen, vascular invasion can and should be detected preoperatively, during the tumor-staging procedure, as it is considered a risk factor for an adverse outcome. Magnetic resonance imaging (MRI) is the only imaging modality that can reliably achieve an accurate diagnosis of vascular invasion (especially extramural), exhibiting significant sensitivity in detecting venous infiltration into the mesorectum [81,84,85]. Recently, MRI was proved to have acceptable sensitivity in detecting EMVI during local staging of colon cancer .
Vascular invasion can also be intramural, as a result of tumor invasion into the submucosa and the wall’s muscular layer . On the other hand, invasion of the extramural veins seems to be associated with the processes taking place at the invasive edge of the tumor, as it is apparently related to the development of TB. Budding cells may independently invade vessels remote from the tumor itself, giving rise to the EMVI . Because of the sequential nature of the tumor development process, intramural and extramural vascular invasion are almost always detected simultaneously .
The clinical significance of intramural vascular invasion is not well established and is consequently controversial, although its presence seems to increase the risk of development of distant metastases . However, EMVI is undeniably a strong predictor of a poor outcome in colorectal cancer patients. It has been repeatedly shown that the presence of EMVI is strongly associated with a greater risk of nodal involvement [80,81], an increase in the incidence of distant metastases, either during diagnosis or during postoperative follow up , a greater recurrence rate  and lower survival rates [80,82,84,87]. Invasion of a large (>3 mm) extramural vein seems to increase the risk for metastases significantly , while the presence of more than four invaded extramural veins is associated with a significant decline in survival rates .
Similarly to TB and PDCs, the presence of EMVI is associated with tumor aggressiveness. As a consequence, detection of EMVI in early-stage colorectal cancer patients might serve as an indicator of the need for administration of chemotherapy . On the other hand, persistent EMVI positivity following neoadjuvant therapy is related to a poor outcome in colorectal cancer patients. In that case, patients may obtain survival benefit from the use of adjuvant therapy, as it has been shown that both the risk of recurrence and DFS are improved following administration of adjuvant chemotherapy  (Tables 1, 2).
PNI provides a distinct way for the colorectal cancer to spread. Having been correlated with an aggressive cancerous phenotype , PNI is a process during which tumor cells invade nervous structures and spread further along nerve sheaths through a complex mechanism that involves a neurotropic factor-mediated tumor-matrix interaction . This process may be greatly facilitated by the presence of budding cells that can infiltrate the myenteric plexus and then spread into the perineurium . Alternatively, it has been proposed that PNI could be the result of stimulation of cancer cell growth by nerves that infiltrate the peritumoral matrix . Extramural PNI has been recognized as a factor that enhances the tumor’s metastatic potential. The clear association between extramural PNI and the presence or development of colorectal cancer liver metastases can be explained by the fact that both the colon and the liver share a sympathetic fiber network of common preganglionic origin, thus allowing the malignant cells to metastasize from the original tumor site toward the liver through sympathetic nerves and without invasion of the surrounding structures .
The true incidence of PNI in colorectal cancer is largely unknown, because it is consistently underreported. Using anti-S100 antibody-based immunohistochemistry techniques, pathologists are now able to diagnose significantly more cases of PNI-positive colorectal cancers, thus enhancing interobserver agreement . The reported incidence varies widely between approximately 7% and 32% [92,95-97] (Table 1).
The deeper the tumor invasion, the greater the number of circumferential resection margins involved, the greater the stage of the disease and/or the lower the tumor differentiation, the higher the incidence of PNI . Tumors located in the retroperitonealized segments (ascending and descending colon cancers) have a higher incidence of PNI, possibly because of their anatomical proximity to the retroperitoneal autonomic nerve plexus [91,93].
Several studies confirm the high probability of lymphovascular invasion and consequently lymph node metastases in PNI-positive tumors [98,99]. Thus, identification of PNI in early stage I/II colorectal cancer may call for changes in the therapeutic approach to those patients, indicating the need for post-surgery chemotherapy [98,100].
The presence of PNI strongly influences the local recurrence rate in patients operated for colorectal cancer. It has been estimated that the 5-year local recurrence rate can be 2.5-fold or even higher in PNI-positive compared with PNI-negative patients . Consequently, adjuvant chemotherapy might be considered as a means of reducing the risk of local recurrence in those patients [96,101].
There is global agreement that the presence of PNI causes a dramatic reduction in survival rates, irrespective of the presence of other adverse factors. Five- or 3-year DFS in patients with PNI-positive tumors has been found to be as low as a quarter of that in patients with PNI-negative colorectal games [90,97,98,101,102]. Likewise, OS has also been shown to be significantly lower in PNI-positive tumors [90,94,95,97,98,101,102]. There is evidence indicating that the presence of PNI in early (stage I/II) colorectal cancer patients reduces survival rates to levels equal to or even below those observed in stage III node-positive colorectal cancer patients .
Although official recommendations cannot be made yet since PNI is not included in the currently used staging systems, administration of adjuvant chemotherapy might be a good option in PNI-positive patients. Postoperative chemotherapy should also be offered in post-neoadjuvant therapy PNI-positive patients, irrespectively of the tumor regression grade achieved  (Tables 1, 2).
TDs have been considered as extramural focal aggregates of cancer cells located in the peritumoral fatty tissue (either mesocolon or mesorectum), which have no continuity with the main tumor mass and are not associated with a lymph node [104,105]. Historically, the exact definition of TDs has changed several times, with AJCC/UICC TNM5 classifying them in the T category as discontinuous tumor extensions if their size is less than 3 mm in diameter [105,106] and AJCC/UICC TNM6 defining them as extratumoral nodules with irregular contours, as opposed to the smooth round surface of the involved regional lymph nodes (N category) [105,107]. Because previous definitions were based on weak and unsubstantiated data , AJCC/UICC TNM7 and TNM8 incorporate TDs into a new N category (N1c), defining them as any cancerous nodule, either microscopic or macroscopic, located in the lymph drainage area of the peritumoral fatty tissue, irrespective of size or shape, as long as there is histologically proven absence of both residual lymphatic tissue in the nodule and regional lymph node metastasis (N0) [105,108,109]. Of course, there is always a possibility for a nodule diagnosed as a TD to be simply a completely infiltrated lymph node in which no normal lymphatic parenchyma remains .
However, there is still argument about what TDs really are, as they usually share different morphology that makes their origin unclear [110,111]. Some believe that TDs simply represent a stage of the lymphovascular and/or perineural invasion process during which malignant cells begin to proliferate, giving rise to distinct nodules of cancer  that have to be distinguished from involved lymph nodes. It has been suggested that the presence of a round shape, lymphoid follicles, peripheral lymphocyte rim, subcapsular sinus, thick capsule and lymph nodes in the surrounding fatty tissue indicates the diagnosis of lymph node rather than TD . There are several reports suggesting different groupings and definitions for TDs, generally classifying them as either lymphovascular, perineural or nodular type [104,105,112,113].
TDs are generally present in about 4.5-45% of colorectal cancer patients [104,105,108,111-118], while their incidence seems to be greater in advanced and/or metastatic tumors [104,105,119] (Table 1). Tumor deposits are more commonly found in colorectal cancers with poor differentiation , involved lymph nodes [104,105,111,119], EMVI [104,111,117], PNI [95,104,115,119], and distant metastases .
The reported incidence varies because of the use of different staging systems during the past two decades. Indeed, it has recently been shown that, in 5th and 6th editions of the AJCC/UICC TNM staging system, the number of involved lymph nodes was significantly overestimated, since many of these nodules were classified as TDs in AJCC/UICC TNM7. As a consequence, the number of cases with TDs and a number of lymph nodes below 12 was significantly increased by the use of TNM7 staging, thus calling into question the accuracy of TNM5/6 lymph node staging . Furthermore, differences in defining TDs have led to significant changes in staging colorectal cancer patients, as the use of different TNM staging systems may cause stage migration of as much as 64% in those with TD [110,121]. Interestingly, despite the fact that TNM7 included the N1c category in an effort to collect enough data to guide future staging systems providing evidence for the significance of TDs, there are currently suggestions about reclassifying TD-positive tumors as metastatic ones (M1) . However, several authors have presented arguments regarding the clinical utility of TNM7 staging, claiming that counting TDs as involved lymph nodes provides better prognostic and predictive value . It is well known that a one-node change in the total number of involved lymph nodes can alter the staging of colorectal cancer disease and may have significant prognostic implications . Nevertheless, the usual practice nowadays is to assess the presence of TDs only in node-negative colorectal cancer patients and not to add the number of TDs to the involved nodes in stage III patients.
Although both are related to poor prognosis, TDs must be distinguished from extracapsular lymph node invasion, which has not yet been accepted as a tumor staging factor . The distinction between them is usually easy to determine since, by definition, the absence of lymph node tissue is a prerequisite for the diagnosis of TDs. As a consequence, TDs and extracapsular lymph node invasion are rarely seen simultaneously .
The presence of TDs has been shown to correlate with an increase in the risk of local recurrence, either before [104,105] or after  administration of neoadjuvant therapy. Detection of TDs, especially those with irregular shape, in stage II colorectal cancers is associated with a higher incidence of local recurrence compared to stage III tumors without TDs .
Both DFS [95,112,115,117-119] and OS [95,105,108,111,117-120] are significantly lower in patients with TD-positive tumors. Moreover, the absolute number and size of TDs seems to affect DFS and OS rates disproportionally [112,122].
Given the great impact of TDs on the prognosis of the colorectal cancer patient, it has been suggested that their detection is necessary for correct staging of the disease, so that TD-positive tumors may be treated with (neo)adjuvant (radio)chemotherapy. Indeed, significant benefits have been reported in early-stage TD-positive colorectal cancer patients, who otherwise would not be offered (radio)chemotherapy . However, there is still argument regarding the clinical use of the presence of TD-like peritumoral nodules following neoadjuvant therapy, as they may simply represent residual tumor microfoci, indicating a good tumor regression grade, rather than real TDs  (Tables 1, 2).
MPs represent aggregations of mucus occasionally seen after neoadjuvant chemoradiotherapy in rectal cancer surgical specimens. The incidence of MPs found in post-chemoradiotherapy surgical specimens from rectal cancer patients varies between approximately 15% and 35% [125-127] (Table 1). Studies referring to both acellular and cellular MPs report that the two types have approximately equal incidence [125,127].
Although the presence of cellular MPs has been reported to correlate with worse tumor differentiation , acellular MPs seem to have no significant impact on the rectal cancer patient’s outcome, provided that complete pathological response to neoadjuvant therapy has been achieved. In fact, neither DFS nor OS differ significantly between rectal cancer patients with or without the presence of MPs in their cancer-free surgical specimens [126-128]. However, detection of viable malignant cells within the MPs is reported to be associated with a 17.8% and 16.9% decrease in 5- and 10-year survivals respectively  (Tables 1, 2).
According to all the above, it is likely that ever-evolving cancer staging systems are nowadays interested in limiting their dependency on the grossly apparent surgical findings to accurately stage colorectal cancer. Instead, they increasingly use histopathologic and immunohistochemical data to precisely detect the severity of the disease and to determine the appropriate treatment accordingly. For that reason, in clinical practice, the identification of TB, PDCs, EMVI, PNI, and TDs needs to be part of the pathological report, since their presence may call for changes in the therapeutic approach. In those cases, clinicians should be aware of the risk of their patients being undertreated if a conventional staging system is used. Accordingly, colorectal cancer patients should be informed about the potential risk of having a poor outcome by choosing minor treatment for an otherwise apparently minor disease.